On Mon, May 10, 2010 at 4:24 PM, Wesley W. Terpstra <wesley at terpstra.ca> wrote: > So gcc optimizes the memcpy away? Interesting/Impressive. Yes. Not surprising, since memcpy of small objects is probably best inlined and unrolled. Also, I suspect that gcc has gotten better at memcpy optimizations as it has taken away union tricks.